A multi-mode coupling damage classification method for composite structures based on acoustic emission signal decomposition.

Multi-mode damage coupling in composite structures is a key factor preventing accurate classification of different damage types. To address this, this paper presents a damage classification framework for composite structures based on acoustic emission (AE) signal decomposition. The approach begins by generating a Peak Frequency-Normalized Count Spectrum using Pearson correlation, principal component analysis, and hierarchical clustering. This spectrum, combined with electron microscopy observations, allows for quick identification of damage types and their frequency ranges, even with limited understanding of damage mechanisms. A customized wavelet packet decomposition filter is then created to decompose AE signals, enabling precise classification of different damage types. To validate the method, multiple tensile tests on adhesive composite joints were conducted, and the AE data were classified using both the proposed method and the K-means method. The results show that, compared to the K-means method, the energy proportions of the three types of damage classified by our method consistently remain in the range of 30%-40%, with the normalized energy proportion of adhesive debonding reaching or even exceeding 50%. Our method more accurately reflects the true damage state of the specimens. It effectively mitigates the negative impact caused by the coupling of multiple damage modes, providing a new perspective for health monitoring of composite structures.